A New Approach to Understanding and Quantifying Storm Surge Predictability

Storm surge is one of the most dangerous hazards of hurricanes; it results in devastating flooding and billions of dollars in damage to coastal regions and is one of the primary hurricane threats for loss of life.  As such it is of great interest to better understand the probability of significant storm surge occurrence and the potential extent of impact at longer lead times to give emergency managers adequate time to plan for necessary evacuation and protective measures.  This work aims to quantify the practical predictability of storm surge at various lead times and the sensitivity of the storm surge to storm parameters such as track, strength, size, and translation speed.  This study also draws a distinction between inundation of a fixed region and inundation following the storm landfall location as the track varies. The latter is not usually considered, but is important for identifying particularly dangerous scenarios within the envelope of possible realizations.  We quantify the predictability of storm surge from both the local and storm-following perspectives.

The ADCIRC model is used to produce an ensemble of storm surge simulations.  The ensemble is generated in an idealized context where the model is driven by best track data and perturbations from the best track (e.g. storm track, maximum wind, storm speed, and storm size).  Inundation metrics are computed for both storm-following inundation and location-based inundation to better understand the predictive nature.  While the magnitude of maximum inundation at a point is often emphasized in storm surge prediction studies, this study focuses on integrated metrics such as inundation volume and spatial extent of inundation along the coast and inland.  Results will be presented from simulations of Hurricane Ike, Hurricane Charley, and a hypothetical storm that combines the size and intensity of Hurricane Charley over the track of Hurricane Ike, to demonstrate the sensitivity of inundation to a certain storm of certain strength as the track varies.  In addition to providing a better understanding of storm surge predictability, results from this study will also be connected to an agent-based model of human response to an impending storm threat as part of a larger project focusing on understanding and improving communication of hurricane risks.